How modify the number of rows and columns of plots, when I plot a sbiopredictionci object?

When the number of groups are so high, using plot with sbiopredictionci object, it plots all the groups left to righ in a single column, is it possible to change this output?.

Thanks

[Florian Augustin]

Hi,

In the current version (R2017b), all inputs to the plot function are plotted in one figure.

If you had a vector of confidence interval results, you could 'down-select' the results you want to plot:

% Example:
fitResults = sbiofit(...);
ciPred     = sbiopredictionci(fitResults);
plot(ciPred([1, 3]));

This does not help, however, if the result returned from sbiopredictionci is not a vector. In this case, you could write a function that copies selected rows/columns from the original plot into a new one. Here is a suggestion on how this could be done:

function selectivePlotCI(ciObj, rowIndices, colIndices)
  ciH = plot(ciObj);
  onCleanup(@()close(ciH)); %#ok<UNONC>
  ax      = reshape(flipud(ciH.Children), ...
                    numel(ciObj(1).ResponseNames), []);
  numCols = numel(colIndices);
  numRows = numel(rowIndices);
  selectedAxes = ax(rowIndices, colIndices);
  for i = 1 : numRows
      for j = 1 : numCols
          subplot(numRows, numCols, (i-1)*numCols+j);
          copyobj(selectedAxes(i, j).Children, gca);
          xlabel(selectedAxes(i, j).XLabel.String);
          if j == 1
              ylabel(ax(rowIndices(i), 1).YLabel.String);
          end
          if i == 1
              title(ax(1, colIndices(j)).Title.String);
          end
      end
  end
end

Just copy/paste/tweak the above function into an m-file named selectivePlotCI.m. Assuming the original plot contains 5 rows and 10 columns, the above function can be used as follows:

fitResults = sbiofit(...);
ciPred     = sbiopredictionci(fitResults);
% Plot rows 3, 5 and columns 1, 3, 10:
selectivePlotCI(ciPred, [3, 5], [1, 3, 10]);

I hope this helps.

-Florian

[Adolfo Garcia Perez]

Thanks Florian, very helpful your answer and your code!!!

[Florian Augustin]

Hi,

starting in R2023b, the internals of the prediction confidence interval plots have changed and the selectivePlotCI function from my previous comment is not working in R2023b or later releases of MATLAB.

Here is an updated version that works in R2023b or later:

function hFig = selectivePlotCI(ciObj, rowIndices, colIndices)

% Plot confidence intervals to get the data.

ciH = plot(ciObj);

ciH.Visible = false;

cleanupObj = onCleanup(@()close(ciH));

ax = reshape(ciH.Children.Children, ...

[], numel(ciObj(1).ResponseNames))';

numCols = numel(colIndices);

numRows = numel(rowIndices);

% Select axes

selectedAxes = ax(rowIndices, colIndices);

% Create a new figure

hFig = figure;

for i = 1 : numRows

for j = 1 : numCols

% Copy confidence interval plot data to new axes:

subAxes = subplot(numRows, numCols, (i-1)*numCols+j);

copyobj(selectedAxes(i, j).Children, subAxes);

xlabel(selectedAxes(i, j).XLabel.String);

if j == 1

ylabel(ax(rowIndices(i), 1).YLabel.String);

end

if i == 1

title(ax(1, colIndices(j)).Title.String);

end

end

end

end

-Florian

[Adolfo Garcia Perez]

Hi Florian,

Thanks for the update

Adolfo